Unraveling the Genotype-Phenotype Landscape of Kleefstra Syndrome Through EHMT1 Variant Analysis

Abstract

Background and Objectives

Kleefstra syndrome type 1 (KLEFS1) is a rare neurodevelopmental disorder caused by pathogenic variants in EHMT1, which play a key role in epigenetic regulation. Core features include global developmental delay (GDD), intellectual disability (ID), facial dysmorphism, congenital heart defects (CHDs), and neuropsychiatric manifestations.

Methods

This retrospective study included patients with KLEFS1 evaluated at Seoul National University Hospital between January 2022 and April 2025. Genetic diagnoses were made using chromosomal microarray or whole-exome or whole-genome sequencing, and clinical data were extracted from medical records. To enhance statistical power, we conducted a joint analysis with established cohorts of at least 30 clinically characterized patients.

Results

Among the 15 patients with KLEFS1, 14 (93.3%) exhibited GDD or ID, with frequent comorbidities such as CHDs (8/12, 66.7%), orthopaedic manifestations (6/12, 50.0%), epilepsy (6/15, 40.0%), and sleep problems (4/11, 36.4%). Brain MRI revealed abnormalities in 10 of the 12 patients (83.3%), including pituitary anomalies in 2 (16.7%). By combining our data with recently published multicenter international cohorts, we assessed genotype-phenotype correlations by comparing the clinical features of 202 patients with different variant subtypes. Patients with multigene deletions had a significantly higher prevalence of CHDs (53.5% vs 25.8%, p value = 2 × 10⁻⁴), cortical visual impairment (12.5% vs 1.7%, p value = 0.004), microcephaly (37.1% vs 17.8%, p value = 0.009), corpus callosum abnormalities (17.9% vs 5.9%, p value = 0.02), and hearing impairment (39.7% vs 25.6%, p value = 0.04) than those with single-nucleotide variants (SNVs) or short insertions/deletions (indels), in addition to the previously identified features including epilepsy, constipation, and short stature. We also observed that a recurrent EHMT1 variant (p.Arg948Trp), identified in multiple unrelated patients, was linked to significant phenotypic heterogeneity, including late-onset neurocognitive regression.

Discussion

Our study showed that patients with multigene deletions are significantly more likely to exhibit multiorgan involvement, including CHDs, cortical visual impairments, microcephaly, corpus callosum abnormalities, and hearing impairments, compared with those with SNVs or indels, emphasizing the importance of genotype-informed management. We report pituitary gland abnormalities in KLEFS1, expanding the phenotypic and neuroimaging spectrum of KLEFS1. The patient showing regression in their 30s necessitates long-term, comprehensive studies to support care.

Introduction

Kleefstra syndrome type 1 (KLEFS1) is a rare neurodevelopmental disorder caused by single-nucleotide variants (SNVs), short insertions/deletions (indels), microdeletions, or structural variants, such as translocations or intragenic duplications of chromosome 9q34.3, resulting in defects of the EHMT1 gene.^1,2 The association between EHMT1 disruption and disease was first described in 2005,³ and more than 200 patients with KLEFS1 have been reported worldwide.^4-6 However, most previous reports have originated primarily from North America and Europe, with limited data available for patients from East Asia.

EHMT1 encodes a histone methyltransferase (euchromatic histone lysine N-methyltransferase 1) that plays a crucial role in epigenetic regulation of gene expression. This enzyme modifies histone proteins by adding methyl groups, which suppresses the activity of specific genes, a process essential for normal development and differentiation.⁷ Disruption of this epigenetic regulation through EHMT1 leads to the characteristic features of KLEFS1.^8,9 The core phenotypic features of KLEFS1 include global developmental delay (GDD), intellectual disability (ID), infantile hypotonia, and distinctive facial dysmorphism. Individuals with KLEFS1 often present with multiple congenital anomalies that affect various organ systems. Cardiovascular problems are particularly prevalent, with 30–40% of patients exhibiting congenital heart defects (CHDs), of which ventricular septal defects are the most common, followed by atrial septal defects.⁴ Microcephaly, skeletal defects, feeding difficulties during infancy, epilepsy, and recurrent infections are also noted.^8,9 Behavioral and neuropsychiatric features are core clinical manifestations of the syndrome, characterized by autism-like traits in childhood followed by regression in adolescence or adulthood.^9-11

The primary aim of this study was to provide a detailed clinical characterization of 15 Korean patients with KLEFS1. Furthermore, to contextualize our findings within the broader clinical spectrum, we conducted a systematic literature review and a joint analysis. This comparative approach was specifically designed to evaluate whether the molecular etiology—namely, multigene deletions vs pathogenic EHMT1 SNVs/indels—significantly influences the clinical phenotype.

Methods

Study Participants

This retrospective study included patients diagnosed with KLEFS1 who visited the Department of Pediatric Neurology and/or Genomic Medicine at Seoul National University Hospital (SNUH) between January 2022 and April 2025. Study participants were identified by querying the electronic medical record system using “Kleefstra syndrome” as the primary search term for diagnosis. The inclusion criteria required molecular confirmation of a 9q34.3 deletion including EHMT1 or a pathogenic EHMT1 (NM_024757.5) variant and the availability of comprehensive medical records. Molecular genetic testing was performed as an integral part of the routine diagnostic workup for each patient in a clinical setting.

Clinical Data Collection and Genetic Analysis

Data on clinical and demographic characteristics were retrospectively obtained from electronic medical records. The collected variables included clinical manifestations, disease course, developmental milestones, and laboratory and neuroimaging results. Chromosomal microarray (Affymetrix CytoScan Dx Assay, Santa Clara, USA) was used to detect large deletions, and SNVs/indels were detected by whole-exome or whole-genome sequencing using the Illumina platforms. Sequence reads were aligned to the human reference genome (hg38) and further processed using the Genome Analysis Toolkit Best Practices workflows. Variant interpretation was performed according to the guidelines of the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP).¹² For further variant analysis, we used in silico predictions, such as phyloP, AlphaMissense, REVEL, Combined Annotation Dependent Depletion (CADD), and SpliceAI, along with public genome databases including ClinVar and gnomAD,^13-15 which were annotated using ANNOVAR.¹⁶ Only variants classified as pathogenic or likely pathogenic were included in subsequent genotype-phenotype correlation analyses. In this study, variants were categorized into multigene deletions and SNVs/indels. The term “multigene deletions” was used specifically to describe large-scale genomic losses involving EHMT1, consistent with established terminology in the Kleefstra syndrome literature. Protein-truncating variants (PTVs) did not include N-terminal truncating variants (N-term), which carry stop codons located within the first 3 exons. Protein-altering variants (PAVs) comprised missense variants located in either the catalytic Su(var)3–9, Enhancer-of-zeste, Trithorax (SET) domain or the ankyrin repeat (ANK) domain. When DNA samples were available, parental testing was performed to assess whether variants of interest were inherited or de novo.

Literature Review

We searched PubMed for English-language articles published between 1990 and 2025 using the queries “EHMT1” OR “Kleefstra.” Articles were first selected based on their abstracts, and the full texts were then reviewed. We focused on cohorts comprising at least 30 patients with genetic confirmation of KLEFS1 and a description of their phenotypes. In addition, patients with detailed explanations of regression with age were included for comparison with our patients. The literature screening process and reasons for study exclusion are presented in eFigure 1.

Joint Analysis and Statistical Methods

Descriptive analyses were performed for our primary cohort (n = 15), followed by a joint analysis to investigate genotype-phenotype correlations. In the joint analysis, we prioritized established large-scale cohorts to ensure the homogeneity of phenotypic descriptions and to avoid the potential for redundant patient reporting across multiple small-scale studies. Consequently, an inclusion threshold of at least 30 patients per study was applied, focusing on cohorts that used standardized clinical evaluation protocols. We further excluded 3 patients who carried large in-frame duplications to maintain cohort homogeneity. This criterion allowed for a more robust and consistent comparative analysis between our patients and the previously reported literature. Categorical variables were compared using the χ² test of independence or Fisher exact tests, as appropriate. Continuous variables were compared using the Kruskal-Wallis test. All statistical analyses were conducted using R software (version 4.5.3, Vienna, Austria). Statistical significance was set at p value <0.05.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the Institutional Review Board of SNUH (2506-096-1649), and the requirement for informed consent was waived because of its retrospective nature.

Data Availability

Deidentified participant data and statistical analysis will be shared on reasonable requests from qualified investigators.

Results

Clinicogenetic Characteristics of Study Patients

Our primary cohort comprised 15 individuals, including 8 male and 7 female patients (Table 1). At the last clinical evaluation, the median age was 10.5 years (range, 3.5–37.8). Notably, the study population consisted predominantly of children (n = 14) aged 11 years or younger, with only 1 adult female patient aged 37.8 years. GDD, ID, or learning difficulties were present in 14 patients (93.3%), highlighting the neurodevelopmental impact of EHMT1 (Table 2 and eTable 1). Other frequently observed clinical features were also identified in our primary cohort. Facial dysmorphism was reported in 10 (66.7%), microcephaly in 7 of 14 patients (50.0%) with available data, and epilepsy in 6 of our patients (40.0%). Scoliosis was also observed in 3 of 10 patients (30.0%), and other orthopaedic problems, including pes planus and pes valgus, were seen in 6 of 12 patients (50.0%).

Table 1.

Clinicogenetic Information of Patients With Kleefstra Syndrome Type 1 in Our Primary Cohort

Patient	Sex/age at last visit	Variant information	ID/GDD	Age at independent walking	Age at first meaningful word	Ophthalmologic abnormality	Hearing loss	Congenital heart defect/cardiac operation	Chest wall deformity	Scoliosis	WM signal change	CC abnormality	Pituitary malformation
Patient 1	F/11	c.673C>T, p.Arg225*, de novo	Y	1.5	Not available	N	Y (conductive, bilateral)	N/N	N	Y	N	N	N
Patient 2	M/11	c.1588C>T, p.Arg530*	Y	1.7	Not available	N	N	Y (ASD)/N	N	Y	Y	N	N
Patient 3	F/37	c.2842C>T, p.Arg948Trp, de novo	N	1.0	1.2	Y (astigmatism, myopia)	N	N/N	N	N	Y	N	N
Patient 4	M/3	c.2842C>T, p.Arg948Trp, de novo	Y	3.3	2.1	N	N	Y (PFO, PS)/N	N	NA	NA	NA	NA
Patient 5	F/7	c.3000del, p.Asp1001Thrfs*9, de novo	Y	1.8	2.8	N	N	N/NA	N	NA	N	N	N
Patient 6	F/10	c.2712+1866G>A, p.?, de novo	Y	1.5	3.0	Y (hyperopia)	N	NA/NA	N	Y	Y	N	Y
Patient 7	F/10	9q34.3 deletion (BAC array and FISH)	Y	Non-A	3.0	Y (bilateral congenital ptosis)	Y (sensorineural, bilateral)	Y (ASD, VSD, PDA)/Y	N	N	NA	NA	NA
Patient 8	M/3	g.136931452_138125937del (1.2 Mb)	Y	Non-A	Non-V	N	N	NA/NA	Y (pectus carinatum)	N	N	N	Y
Patient 9	M/5	g.137540333_138125937del (0.6 Mb)	Y	1.5	1.4	Y (strabismus)	N	NA/NA	N	NA	NA	NA	NA
Patient 10	F/10	g.137044229_137843854del (0.8 Mb)	Y	2.3	1.9	N	N	Y (ASD)/NA	N	N	Y	Y	N
Patient 11	M/11	g.136668488_138220891del (1.6 Mb)	Y	3.8	3.8	N	Y (sensorineural, unilateral)	Y (PFO)/NA	N	N	N	Y	N
Patient 12	M/6	g.136468518_138059695del (1.6 Mb)	Y	2.5	Non-V	Y (hyperopia, strabismus)	Y (sensorineural, bilateral)	Y (ASD, VSD, PDA)/N	N	N	N	Y	N
Patient 13	M/11	g.135296424_138111185del (2.8 Mb)	Y	6.0	Not available	Y (hyperopia)	N	N/N	Y (pectus excavatum)	NA	Y	N	N
Patient 14	M/10	g.135282597_138124196del (2.8 Mb)	Y	Ambulation with assist at 9	5.8	Y (hyperopia, strabismus)	Y (sensorineural, bilateral)	Y (SD)/N	Y (pectus excavatum)	NA	Y	N	N
Patient 15	F/7	g.135179450_138134124del (3.0 Mb)	Y	Non-A	Non-V	Y (strabismus)	N	Y (VSD, ASD, PDA)/Y	N	N	N	N	N

Abbreviations: ASD = atrial septal defect; BAC = bacterial artificial chromosome array; CC = corpus callosum; F = female; FISH = fluorescence in situ hybridization; GDD = global developmental delay; ID = intellectual disability; M = male; N = no; NA = not applicable; Non-A: nonambulatory at last visit; Non-V = nonverbal at last visit; PDA = patent ductus arteriosus; PFO = patent foramen ovale; PS = pulmonary stenosis; SD = septal defect; VSD: ventricular septal defect; WM = white matter; Y = yes; yr = year.

Table 2.

Comparison of Manifestations in Patients With SNVs/Indels and Multigene Deletion in Our Primary Cohort

	Total cohort (n = 15)	Multigene deletion (n = 9)	SNVs/indels (n = 6)
Sex, male (%)	53.3 (8/15)	66.7 (6/9)	33.3 (2/6)
Age at last visit, yr, median (min-max)	10.5 (3.5–37.8)	10.5 (3.6–11.9)	10.7 (3.5–37.8)
Facial dysmorphism (%)	66.7 (10/15)	88.9 (8/9)	33.3 (2/6)
Nervous system (%)
ID/GDD	93.3 (14/15)	100.0 (9/9)	83.3 (5/6)
Age at independent walking, yr, median (min-max)a	1.8 (1.0–6.0)	2.5 (1.5–6.0)	1.6 (1.0–3.3)
Epilepsy	40.0 (6/15)	44.4 (4/9)	33.3 (2/6)
Microcephaly	50.0 (7/14)	44.4 (4/9)	60.0 (3/5)
Sleep problem	36.4 (4/11)	0.0 (0/5)	66.7 (4/6)
Regression	6.7 (1/15)	0.0 (0/9)	16.7 (1/6)
Brain imaging abnormality (%)	83.3 (10/12)	100.0 (7/7)	60.0 (3/5)
White matter signal change	50.0 (6/12)	42.9 (3/7)	60.0 (3/5)
Corpus callosum abnormality	25.0 (3/12)	42.9 (3/7)	0.0 (0/5)
Pituitary abnormality	16.7 (2/12)	14.3 (1/7)	20.0 (1/5)
Sensory system (%)
Hearing impairment	33.3 (5/15)	44.4 (4/9)	16.7 (1/6)
Digestive system (%)
Feeding difficulties during infancy	25.0 (2/8)	40.0 (2/5)	0.0 (0/3)
Constipation	44.4 (4/9)	40.0 (2/5)	50.0 (2/4)
Cardiovascular system (%)
Congenital heart defect	66.7 (8/12)	85.7 (6/7)	40.0 (2/5)
Cardiac operation	25.0 (2/8)	33.3 (2/6)	0.0 (0/2)
Musculoskeletal system (%)
Scoliosis	30.0 (3/10)	0.0 (0/6)	75.0 (3/4)
Other orthopaedic problem	50.0 (6/12)	50.0 (4/8)	50.0 (2/4)
Endocrine system (%)
Hypothyroidism	22.2 (2/9)	20.0 (1/5)	25.0 (1/4)
Short stature	26.7 (4/15)	44.4 (4/9)	0.0 (0/6)
Obesity	6.7 (1/15)	11.1 (1/9)	0.0 (0/6)
Immune system (%)
Recurrent infections	13.3 (2/15)	11.1 (1/9)	16.7 (1/6)

Abbreviations: GDD = global developmental delay; ID = intellectual disability; max = maximum; min = minimum; yr = year.

^aData were collected where available.

Genotype-Phenotype Correlations in the Primary Cohort

Six patients had pathogenic or likely pathogenic SNVs or indels, and 9 had multigene deletions encompassing EHMT1. Among the SNVs or indels, 3 were truncating variants and 2 were missense variants affecting the ANK domain, all of which have been previously reported in ClinVar except 1 deep intronic variant (c.2712 + 1866G > A). The Sanger test on the trio confirmed that the c.2712 + 1866G > A variant arose de novo in Patient 6 (eFigure 2), whose biological parentage was verified via trio whole-exome sequencing. This variant was predicted to have a deleterious effect on splicing by the splicing prediction tool SpliceAI (delta score: 0.80) and was classified as likely pathogenic. Clinically, Patient 6 exhibited features highly consistent with Kleefstra syndrome, including neurodevelopmental delay, severe ID, and characteristic dysmorphic facies such as a sunken midface and hypertelorism, further supporting the diagnosis. Except for c.2842C > T (p.Arg948Trp), all missense variants found in our primary cohort and reported by Rots et al. and Frazier et al. had an allele frequency of 0 in gnomAD (v.4.1.0), the public variant database for the general population.¹⁵ On the contrary, the allele frequency of the c.2842C > T (p.Arg948Trp) variant was 6.57 × 10⁻⁶ in the overall gnomAD genome, and the person was aged between 55 and 60 years (eTable 2).¹⁵

CHDs were commonly observed in our patients, some of whom required treatment. As shown in Table 2, 8 of 12 patients with KLEFS1 (66.7%) had CHD. Specifically, 2 of 5 patients with SNVs or indels (40.0%) and 6 of 7 patients with multigene deletions (85.7%) had CHDs. Surgical intervention for CHDs was necessary in 2 patients (25.0% of those with CHDs), both of whom had multigene deletions.

Brain imaging of patients with KLEFS1 has revealed various structural abnormalities. Brain MRI was performed in 12 of the 15 patients, with 5 undergoing serial imaging follow-ups, which demonstrated no interval changes in the findings (eFigure 3). Abnormal findings were observed in 10 of 12 patients (83.3%). The most frequently observed finding was T2 signal hyperintensity in the cerebral white matter (WM) including peritrigonal area, affecting 6 of 12 patients (50.0%) (Figure 1A). Additional features included corpus callosum abnormalities, including dysplastic corpus callosum (Figure 1B) in 3 of 12 (25.0%) and pituitary abnormalities in 2 patients (16.7%). One patient (Patient 8) demonstrated the absence of the neurohypophysis, and another (Patient 6) had small anterior pituitary gland (Figure 1C). Despite the structural abnormalities, both patients showed normal pituitary hormone levels at ages 2 and 5 years, respectively.

Figure 1. Brain MRI Features Among Variant Types in Patients With Kleefstra Syndrome Type 1.

(A) T2 axial image of Patient 2 at age 7 showing T2 high signal intensities of the peritrigonal white matter (WM) (arrowheads). (B) T1 sagittal image of Patient 10 at the age of 2 years with a dysplastic corpus callosum (arrowhead). (C) Patient 6 at the age of 4 years showed an empty pituitary gland (arrowhead). (D) Comparison of brain MRI features among patients with different genotypes from our primary cohort and literature review. Indel = short insertion/deletion; SNV = single-nucleotide variant.

Brain imaging abnormalities were more prevalent in patients with multigene deletions; all 7 imaged patients revealed structural abnormalities. All 3 patients with corpus callosum abnormalities had multigene deletions. Figure 1D and eTable 3 present the results of a joint analysis of our primary cohort and previously published articles with available brain MRI descriptions. The results show that corpus callosum abnormalities were more frequent in the multigene deletion group than those with SNVs/indels (26.9% vs 12.9%, p value = 0.02).^8,17

Joint Analysis With Previously Reported Cohorts

Our primary cohort data were jointly analyzed with previously reported cohorts for more comprehensive phenotype-genotype correlation of KLEFS1.^8,9 The cohort by Rots et al. comprised 209 patients from 200 families, recruited through local Kleefstra syndrome expertise centers and international collaborators, of whom 122 patients with available clinical and appropriate variant information were included in the joint analysis. The cohort by Frazier et al. consisted of 65 patients evaluated at the Boston Children's Hospital Kleefstra Clinic. A total of 202 patients were included in the joint analysis, comprising 129 individuals with EHMT1 SNVs or indels and 73 with multigene deletions (eFigure 4). The multigene deletion group consisted of patients from our primary cohort (n = 9) along with those reported by Rots et al. (n = 34) and Frazier et al. (n = 30). For the SNV/indel group, data were derived from our primary cohort (n = 6) and the cohorts reported by Rots et al. (n = 88) and Frazier et al. (n = 35) (Figures 2 and 3, and eTable 4). Several features, including epilepsy (33.3% vs 7.3%, p value = 0.005), short stature (23.7% vs 6.0%, p value = 0.01), and constipation (52.2% vs 37.0%, p value = 0.049), were significantly more prevalent in the multigene deletion group (eTable 4). In particular, we showed a significant association between CHDs and multigene deletion, which were not seen in previous cohorts (53.5% vs 25.8%, p value = 2 × 10⁻⁴). Similarly, multigene deletions were also significantly associated with cortical visual impairment (CVI) (12.5% vs 1.7%, p value = 0.004), microcephaly (37.1% vs 17.8%, p value = 0.009), and hearing impairment (39.7% vs 25.6%, p value = 0.04), associations that have not been previously reported. Conversely, the previously observed higher prevalence of GDD and/or ID in patients with multigene deletions compared with those with SNVs or indels was not maintained in the joint analysis.^5,6 Following the classification previously reported⁸ and combining PAVs in ANK and SET domains, variants were grouped into 4 categories: multigene deletions, PAVs, PTVs, and N-term. Notably, our analysis revealed that patients with PTVs (98.0% vs 62.5%, p value = 0.006) and PAVs (94.3% vs 62.5%, p value = 0.04) had a significantly higher prevalence of ID/GDD compared with those with N-term (eTable 5, eTable 6, and eFigure 5). Patients with PTVs have more frequent recurrent infections compared with those with PAVs (63.6% vs 29.0%, p value = 0.01). Although not statistically significant, a higher prevalence of non-CNS manifestations, including hearing impairment, constipation, and gastroesophageal reflux disease, was observed in patients with PTVs compared with those with PAVs.

Figure 2. Comparison of Clinical Manifestations in Patients With Kleefstra Syndrome Type 1 According to Variant Types.

Patients with multigene deletion had significantly higher frequencies of epilepsy, microcephaly, hearing impairment, cortical visual impairment, constipation, congenital heart defects, and short stature than those with SNVs or indels. *, p value < 0.05; **, p value < 0.01; ***, p value < 0.001. ※ Includes data from our primary cohort and the cohorts reported by Rots et al. (2024) and Frazier et al. (2025). GDD = global developmental delay; ID = intellectual disability; indel = short insertion/deletion; SNV = single-nucleotide variant.

Figure 3. Comparison of Clinical Manifestations in Patients With Kleefstra Syndrome Type 1 According to Variant Types in Our Study and Previous Studies.

The comparison of clinical features across our primary cohort and the cohorts reported by Rots et al. (2024) and Frazier et al. (2025) shows that intellectual disability and global developmental delay were highly prevalent in both multigene deletion and SNV/indel groups across all cohorts, while other features show variable prevalence across cohorts and genotypes. GDD = global developmental delay; ID = intellectual disability; indel = short insertion/deletion; SNV = single-nucleotide variant.

Patients With the Recurrent Missense Variant p.Arg948Trp

A recurrent missense variant, EHMT1 p.Arg948Trp, was observed in 2 unrelated patients in our primary cohort and in 3 additional cases from previous literature (Table 3). Despite sharing the same genetic alterations, phenotypic expressions vary. Patient 4 in our primary cohort and the 3 individuals previously reported⁸ all exhibited GDD and/or ID, with 2 showing mild ID on formal cognitive testing. By contrast, Patient 3 in our primary cohort presented with a markedly different clinical course, characterized by normal development through young adulthood and the onset of seizures at the age of 28 years. In her early 30s, she began to experience progressive neurocognitive regression, which was associated in time with frequent nocturnal awakenings and cognitive decline. Considering the heterogeneous clinical presentations and the occurrence of this variant in 1 individual from gnomAD, it is plausible that this missense variant contributes to clinical variability.

Table 3.

Comparison of Patients With p.Arg948Trp Missense Variant

Patient ID	Age range at last visit	ID/GDD (IQ, if data exists)	Age at first meaningful word	Regression	Epilepsy	Microcephaly	Ophthalmologic problem	Congenital heart defect	Gastrointestinal problem	Infantile hypotonia	Neuropsychiatric disease
Patient 3	4th decade	N	1.2	Y	Y	Y	Y (astigmatism, myopia)	N	N	N	Y (psychosis)
Patient 4	Toddler	Y	2.1	N	N	NA	N	Y (PFO, PS)	N	N	N
Patient Aa	Toddler	Y (68)	2.3	N	N	N	Y (strabismus)	NA	Y	Y	N
Patient Bb	Adolescent	Y (57)	NA	N	N	N	Y (strabismus)	Y	N	N	Y (ADHD)
Patient Cc	Toddler	Y	NA	N	N	N	Y (retinoblastomad)	NA	N	N	N

Abbreviations: ADHD = attention-deficit hyperactivity disorder; F = female; GDD = global developmental delay; ID = intellectual disability; M = male; N = no; NA = not applicable; PFO = patent foramen ovale; PS = pulmonary stenosis; WM = white matter; Y = yes; yr = year.

^aPatient A, KS110043, VUSM110019 (Rots et al.).

^bPatient B, VUSM110018 (Rots et al.).

^cPatient C, KS110105 (Rots et al.).

^dThe patient had a pathogenic RB1 variant, explaining the ophthalmologic problem observed in this patient.

Regression in Subset of Patients With KLEFS1

A joint analysis of our patients with cohorts reported by Rots et al. and Frazier et al. found that 22.7% of patients with SNVs or indels and 31.8% of patients with multigene deletions exhibited regression (Figure 3 and eTable 4). However, the incidence did not differ significantly between the multigene deletion and SNV/indel groups (p value = 0.22). The observed regression in Patient 3 was characterized by a loss of previously acquired skills in multiple domains, including self-care, independent living, social interaction, and fine motor coordination. In addition to Patient 3 in our primary cohort, 5 other patients with detailed descriptions of regression including the language domain, beginning between the 2nd and 4th decades, were identified through a literature review (Table 4). Similarly, this pattern was not specific to any particular variant type; 3 patients had multigene deletions, 1 had a small deletion, 1 had a splice site variant, and our patient had a missense variant.^10,11 Notably, 2 (Patients 3 and F) achieved typical developmental milestones during early life, including acquiring bilingual language skills and engaging in sports and music, suggesting relatively preserved early neurocognitive function. However, between 2^nd and 4^th decade, regression emerged, coinciding with the onset of sleep disturbances. Four patients (Patients 3, E, F, and G) among the 6 with a detailed clinical course of regression had sleep-wake cycle problems, such as frequent nocturnal waking and an abnormal sleep-wake cycle, during the year when regression became evident.

Table 4.

Comparison of Patients With Kleefstra Syndrome Experiencing Regression

	Our primary cohort	Verhoeven et al., 2011			Verhoeven et al., 2010
Patient IDa	Patient 3	Patient D	Patient E	Patient F	Patient G	Patient H
Age range at last visit	4th decade	5th decade	4th decade	2nd decade	6th decade	6th decade
Variant information	c.2842C>T, p.Arg948Trp, de novo	c.3087+1G>T, de novo	c.3181-80_3233del, de novo	9q subtelomeric deletion	9q subtelomeric deletion	9q subtelomeric deletion
Perinatal history	None	Feeding problem, growth retardation	None	SGA, microcephaly, VSD	None	Feeding problem
Regression onset age, yr	32	38	15	19	20	Late teens
Maximum output/activity	Professional musician	Receptive language of 3.5 yr	In early 10s: cycling, swimming, skating	Gymnastic, bilingual speech	—	—
1st decade (0–9 yr)	Normal development	GDD, decreased social interaction, emotional instability	—	VSD, sleep problem	1 yr: suspicious GDD, microcephaly, facial dysmorphism	24mo: independent walking; preschool: 1st simple words; 6y: febrile seizure; 9y: severe learning disability
2nd decade (10–19 yr)	—	—	15 yr: permanent institutionalization, bruxism, complete loss of speech, abnormal sleep-wake cycle	Anxiety, irritability 19 yr: global function deterioration, speech and language regression, regular nocturnal awakening	10 yr: hyperkinesia, stereotypy, rigidity, slight hypotonia; 16 yr: tonic-clonic, absence seizure	16 yr: FSIQ 30; 19 yr: admission to a day activity center
3rd decade (20–29 yr)	28 yr: seizures, staring events	—	—	NA	20 yr: fine-motor dyspraxia hyperactivity, aggressive outburst, orofacial stereotypies, frequent nocturnal activities, difficulty falling asleep	29 yr: permanently institutionalized, frequent myoclonic twitches; hypokinesia; rigid movements; paroxysmal temper tantrums
4th decade (30–39 yr)	32 yr: cognitive decline, hand stereotypy, regular awakening at night (2–3 times/night)	38 yr: decreased global function, impulsivity, temper tantrums, rhythmic jerks	33 yr: estimated IQ: 30; stereotyped movement, pseudospastic walking pattern	NA	—	—
5th decade (40–49 yr)	NA	43 yr: abnormal posture—both arm flexure, both hands twisted, thumbs rotated inward	NA	NA	—	45 yr: severe regurgitation 46 yr: rumination, stereotyped tongue movements
6th decade (50–59 yr)	NA	NA	NA	NA	53 yr: estimated IQ, 30; extreme hypoactivity, lack of spontaneous movements, both arm flexure, twisted hand posture; nonverbal	59 yr: estimated IQ, 25–35; orofacial and hand stereotypy, staring, rigid walking, lack of spontaneous movements

Abbreviations: F = female; FSIQ = full scale IQ; GDD = global developmental delay; M = male; mo = month; NA = not available because the patient had not reached the required age; SGA = small for gestational age; VSD = ventricular septal defect; yr = year.

^aPatients D, E, and F (Pts 1, 2, and 3 in the study by Verhoeven et al., 2011, respectively); Patients G and H (Pts A and B in the study by Verhoeven et al., 2009, respectively).

Discussion

Our study highlights distinct genotype-phenotype correlations in patients with KLEFS1, particularly between those with EHMT1 SNVs or indels and those with multigene deletions. Several clinical features in line with previous studies, such as epilepsy, short stature, and constipation, were more commonly observed in the multigene deletion group.^8,9 In addition, we were able to show a significant relationship between multigene deletion and CHD, CVI, microcephaly, abnormal corpus callosum structure, and hearing impairment, which was not achieved in previous studies. In addition, our study showed highly prevalent (83.3%) brain structural abnormalities in patients with KLEFS1, including 2 patients with structural abnormalities of the pituitary gland. While EHMT1 has been primarily characterized by its enzymatic “writer” function, our understanding of its role has evolved. Emerging evidence suggests that its role as a methyltransferase may be dispensable in certain biological contexts, with its noncatalytic scaffolding roles playing a more prominent part in the pathogenesis of Kleefstra syndrome.^8,18 This functional complexity may further explain the phenotypic diversity observed in patients with different types of variants.

The neuroimaging findings in our cohort are consistent with those reported in the literature and offer new insights. A literature review showed approximately 25% of patients with KLEFS1 exhibit normal brain MRI findings.¹⁷ Another review¹⁹ showed that 58%–63% of patients with KLEFS1 exhibited abnormalities on brain imaging, depending on variant type. Commonly reported abnormalities include cerebral atrophy, periventricular leukomalacia, hypoplasia of the inferior cerebellar vermis, WM signal changes, and agenesis or dysgenesis of the corpus callosum. Among 8 patients with KLEFS1, all showed abnormal corpus callosal structure; 6 had hypoplasia, 1 had dysgenesis, and another 1 had agenesis.²⁰ In our study, 83.3% of patients who underwent brain imaging demonstrated abnormal findings, with WM changes and corpus callosum abnormalities being the most prevalent. In addition, we report 2 patients with structural pituitary gland anomalies, one with a small anterior pituitary gland and the other with an absent neurohypophysis. While a previous case report described pituitary tethering, imaging was not provided.²¹

The missense variant p.Arg948Trp, located in the ANK domain, has been reported multiple times. This variant showed high CADD and AlphaMissense scores, 23.1 and 0.9936, respectively, supporting its predicted pathogenicity (eTable 2). Functional investigations demonstrated that p.Arg948Trp reduces histone H3 tail peptide binding and protein thermostability in vitro.⁸ However, interactions with EHMT2 and overall histone methyltransferase activity were unaffected, and cellular stability in vivo remained unchanged. Likewise, although the p.Arg948Trp variant is associated with a KLEFS1-specific DNA methylation signature, patients carrying this variant often exhibit a relatively milder phenotype. Taken together, these findings may reflect the phenotypic variability identified in this study; while some patients presented with GDD and mild ID, 1 individual (Patient 3) had minimal symptoms until adulthood, emphasizing the possibility of genetic or environmental modifiers influencing phenotypic expression in EHMT1-related disorders. Remarkably, 1 individual was identified with the variant in gnomAD, suggesting incomplete penetrance, late-onset symptoms, or a milder phenotype.

Regression appears to affect a subset of patients. The reported regression prevalence was approximately 11%, including a young adult with progressive cognitive decline and cerebral atrophy.⁹ Other cases with regression exhibited disrupted sleep, apathy, and catatonia-like symptoms characterized by stereotypy, stupor, and staring episodes.^22,23 Similarly, 2 adult patients with moderate ID underwent regression in adulthood, displaying alternating periods of apathy and catatonic behavior.¹⁹ In our cohort and through the review of previous cases, we identified 2 distinct patterns: one group exhibited significant developmental delay from infancy or childhood, while another showed near-normal early development, before regressing in their late teens to early 30s. This bimodal onset suggests that EHMT1 may play critical roles not only in early neurodevelopment but also in late adolescence or early adulthood. Analysis of the BrainSpan RNA expression profiles revealed a decrease in EHMT1 expression from the fetal period to early adolescence, a second increase from late adolescence through the third decade of life, and a decline thereafter (eFigure 6).²⁴ The elevated expression during the fetal period suggests a critical role for EHMT1 in early brain development. The modest elevation in expression during late adolescence and young adulthood may suggest a period of enhanced functional relevance, which could contribute to the regression reported in individuals in this age group. Further research is needed to elucidate the molecular mechanisms underlying the regression and the symptom onset of KLEFS1. As given in Table 4, patients with regression experienced sleep disturbance within the year. However, given the wide variability in reported prevalence of sleep problems in KLEFS1, ranging from 2.9% to 63%, it remains unclear whether sleep disturbances are related to the onset of regression or are merely coincidental.^5,6,8,9 Notably, our finding aligns with a previous report that suggested sleep disturbances may precede regression episodes, highlighting a potential temporal or prodromal association that warrants further investigation.

This study contributes to the limited body of literature on KLEFS1 by presenting a well-characterized cohort that includes both EHMT1 SNVs or indels and multigene deletions, as well as neuroimaging features. However, several clinical features varied in prevalence among 3 cohorts. Our KLEFS1 study showed a higher prevalence of CHD, and the cohort reported by Frazier et al.⁹ had a high prevalence of hearing impairment. The cohort reported by Rots et al.⁸ showed a higher prevalence of recurrent infection; behavioral and psychiatric disorders, including autism spectrum disorders; and regression. A potential bias related to the age at clinical presentation cannot be entirely ruled out because individual-level age data from the joint analysis cohorts were not sufficiently available to allow for a formal age adjustment. The notable differences in the reported prevalence of specific clinical features may arise from different evaluators or study populations. The 2 cohorts and many caregiver-reported studies have predominantly originated from North America and Europe.^5,6,8,9,19 Our cohort broadens the scope by incorporating patients from East Asia; however, further studies are required to determine whether there are ethnicity-specific differences.

Recently, the European Reference Network (ERN) Ithaca Guideline Working Group published the first international, evidence-based clinical guideline for KLEFS1, comprising 66 consensus-based recommendations spanning diagnosis, neurodevelopment, mental health, sensory, cardiac, and family support domains across the lifespan.²⁵ While this guideline establishes a uniform minimum standard of care, its recommendations are largely genotype-independent and derived predominantly from North American and European cohorts. Our study complements this framework in several respects. First, by demonstrating that multigene deletions confer a significantly higher burden of CVI and hearing impairment compared with SNVs/indels, our findings support a genotype-informed approach to implementing the guideline's screening recommendations, suggesting that patients with multigene deletions may warrant closer tailored ophthalmologic and audiologic surveillance. Second, our detailed characterization of the bimodal pattern of regression, including its temporal association with sleep disturbance and correlative EHMT1 expression data, provides natural history evidence that may help refine the guideline's recommendation for “urgent specialist assessment” of regression into more specific, age-stratified monitoring protocols. Third, our identification of structural pituitary anomalies expands the neuroimaging phenotype beyond what is currently addressed in the guideline's physical health domain. Finally, as one of the few reported KLEFS1 cohorts from East Asia, our data contribute toward assessing the generalizability of these internationally derived recommendations across diverse populations.

This study has some limitations, such as the limited number of patients in our cohort. As a retrospective chart review, inconsistencies or omissions might have occurred in the clinical documentation. However, we were able to expand the clinical and neuroimaging spectrum of KLEFS1 by comparing patients with EHMT1 multigene deletions and those with SNVs or indels, revealing key differences in phenotypic severity and specific clinical features. We highlight the importance of genotype-informed evaluation, particularly in the context of CHDs, CVI, microcephaly, corpus callosum abnormalities, and hearing impairment, along with epilepsy, short stature, and constipation. The phenotypic variability observed even among patients with the same missense variant stresses the likely presence of genetic or environmental modifiers that influence disease expression, emphasizing the need for comprehensive long-term clinical monitoring of a larger number of patients with KLEFS1.

Glossary

ANK

ankyrin repeat

CADD

Combined Annotation Dependent Depletion

CHD

congenital heart defect

CVI

cortical visual impairment

GDD

global developmental delay

ID

intellectual disability

KLEFS1

Kleefstra syndrome type 1

PAV

protein-altering variant

PTV

protein-truncating variant

SNUH

Seoul National University Hospital

SNV

single-nucleotide variant

WM

white matter

Author Contributions

J.M. Kim: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data. J.H. Cha: drafting/revision of the manuscript for content, including medical writing for content. W. Kim: drafting/revision of the manuscript for content, including medical writing for content. S. Y. Kim: drafting/revision of the manuscript for content, including medical writing for content. J.M. Ko: major role in the acquisition of data. B. Lim: major role in the acquisition of data. S. Lee: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. J-H. Chae: major role in the acquisition of data; study concept or design.

Study Funding

This research was supported and funded by the SNUH Lee Kun-hee Child Cancer & Rare Disease Project, Republic of Korea (grant number: 22B-001-0100). This research was supported by a grant of Korean ARPA-H Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: RS-2025-25454991).

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/NG.